Evaluation of an expert system for fault detection, isolation, and recovery in the manned maneuvering unit
John Rushby, Judith Crow · 1990
We explore issues in the specification, verification, and validation of AI-based software using a prototype Fault Detection, Isolation, and Recov-ery (FDII_) system for the Manned-Maneuvering Unit (MMU). We use the MMU FDIR system, which is implemented in CLIPS, as a vehicle for ex-ploring issues in the semantics of CLIPS-style, rule-based languages, the verification of properties relating to safety and reliability, and the static and dynamic analysis of knowledge-based systems. Our analysis reveals errors and shortcomings in the MMU FDII _ system and raises a number of issues concerning software engineering in CLIPS. In the course of this work we came to realize that the MMU FDIR system does not conform to conventional definitions of AI software, despite the fact that it was intended and indeed presented as an AI system. We discuss this apparent disparity and related questions such as the role of AI techniques in space and aircraft operations and the suitability of rule-based languages